University of Houston

From the SelectedWorks of L. A. Witt

2011

Assessing the cost of underperformance: A

computer programmer example

Available at: http://works.bepress.com/witt/30/

 Assessing the Cost of Underperformance:
A Computer Programmer Example
Rodger W. Griffeth
Ohio University
L.A. Witt
University of Houston

Carlos Polk
Civil Service, Federal Government
Sean Robinson
Ohio University

Rebecca Thacker
Ohio University

Kori Callison
University of Houston

Conversations with our SIOP colleagues suggest that many, if not most.
of us find considerably more support from executives than managers for our
projects. Whereas executives often take a long-term view that accepts short-
tenn costs as necessary for long-tenn gain, managers tend to focus on short- j
:,
tenn costs. It appears that it is quite common for us to face resistance from
line managers when we are seeking their support (i.e., funding, permission to
collect data from employees during work hours, or both) for an HR study that
is intended to help them! Unlike corporate attorneys whose advice managers
seem to follow without question, we typically find ourselves jumping through
hoop after hoop to "justify" what we propose. We offer an approach to reduce
the number of these hoops.
What do managers care about? The pay of line managers is typically a
function of the extent to which they meet their business unit goals, which are
most often operationalized in terms of short-term (i.e., quarterly or annual)
financial petformance (i.e., adherence to their allocated budget) and opera-
tional perfonnance (i.e., production in the fonn of revenues, service, and/or
goods). As a consequence, managers think in the short term and tend to care
most about meeting operational objectives and not spending too much
money. This requires a delicate balancing act, as business conditions some-
times require allocating unbudgeted monies {e.g., overtime pay) to maintain
operational petfonnance levels. The bottom line, though, is that managers are
expected to meet expected operational performance levels. Anecdotal evi-
dence suggests that executives are more tolerant of managers who go slight-
ly over budget to meet operational performance goals than of managers who
stay well within the budget but fall short of operational performance goals.

The Industrial-Organizational Psychologist 13

 Given that managers tend to care most about meeting operational objec-
tives and not spending too much money, it is not surprising that HR-related
projects are among the first to get cut when budget constraints are identified.
That is, "desired but not required" HR projects are typically seen as expens-
es designed to "improve" operations. Perhaps consistent with the old adage,
"if it ain ' t broke, don't fix it,'' when managers are faced with the need to hold
the line on expenses and/or maintain production levels, they likely view HR-
related projects as risky from an return on investment (ROf) perspective. It is
not surprising that our efforts to position projects as high utility are met with
resistance because they threaten what managers care most about: short-tenn
operational (i.e., taking employees offiine to participate in a study) and finan-
cial perfonnance (i.e.. funding the study).
Work on attitude change suggests that persuasive attempts are "likely to be
effective to the extent that the function of, or reason for holding, the position
outlined in the appeal matches the function underlying recipients attitude"
(Wood. 2000, p. 545). To us. that means that we need to position what we do
in tenns of what managers care most about: short-term operational and finan-
cial performance. Consequently, we advocate positioning the "underperfor-
mance" of employees as a threat to what managers care about. As a threat,
"underperformance" may have multiple causes. including dysfunctional selec-
tion systems, poor leadership, and inadequate training. Hence, it may be appro-
priate as a universally relevant label for the threats that we typically can
address. We know from physician appeal research-work intended to ascertain
how physicians can most effectively persuade patients to discontinue health-
damaging habits--that well-defined threats presented with well-defined solu-
tions arc more effective than well-defined opportunities for better health in gen-
eral (e.g., Prentice-Dunn, Jones, & Floyd, 1997; Rogers, 1983). In other words.
as even most lay observers of political campaigns well appreciate, "going neg-
ative" works. Consistent with work by Macan and Foster (2004), we argue that
this approach is more likely to work with managers than appeals based on
opportunities to improve performance over the long term. As many managers
find themselves "too busy fighting alligators to drain the swamp," it behooves
us to position our projects in tenns of a solution to a short-tenn threat. Doing
so moves our projects from the "desired'' to "required" column.
What evidence can we offer to identify a threat? We have for many years
applied various fonns ofutility analysis to demonstrate the effectiveness of HR
practices. For example, practitioners have presented the number of employees
affected x quality of the HR program/practice, minus the costs (Boudreau,
1991; Cascio, I 991 ). Whereas managers apparently appreciate utility analysis
and the light it can shed on financial benefits from HR practices (Carson, Beck-
er, & Henderson, 1998), evidence suggests that providing the results of a utili-
ty analysis to managers does not enhance the level of acceptance of HR prac-
tices (Latham & Whyte, 1994; Whyte & Latham, 1997; Carson et al., 1998).

14 April 2011 Volume 48 Number 4

 We suspect that the primary reason that utility analysis results have been less
than convincing is not that utility analysis comes across as complex (Carson et
al., 1998). Rather, we argue that managers may be predisposed to keep HR proj-
ects in the low-priority category because they see threats (i.e., other issues) to
their performance goals as clear priorities. In other words. we often have not
made the sale that managers need to address underperfonnance because it is a
threat to operational (and sometimes also financial) performance goals.
We offer a simple example of how one might apply utility analysis to
identify the cost of underperfonnance. We discuss a study in which we com-
puted the costs of underperfonnance across computer programmer/analysts
in a software development organization. In tough economic times. organiza-
tions can ill afford to absorb the costs ofunderperfonning employees. Rather
than thinking of the value a particular HR practice or program adds to the
organization, why not calculate the cost of keeping employees whose per-
formance is not up to par, a threat to the success of managers?
To calculate the costs of underperforrnance in this study, we applied the
Cascio-Ramos estimate of performance in dollars (CREPID; Cascio &
Ramos, 1986). The rationale underlying CREPID assumes that an "organiza-
tion's compensation program reflects current market rates for jobs, then the
economic value of each employee's labor is reflected in his or her annual wage
or salary'' (Cascio. 1991. p. 213-214). Interpreting the results of our calcula-
tions presented here requires acceptance of some underlying ac;sumptions. The
first assumption is that there is an outcome against which performance level
can be measured. Available operationalizations for undcrperformance include
involuntary tenninations, product flaws, customer complaints. employee
errors. and schedule failures, all of which are direct and relatively easy to com-
pute. Other measures include costs of absence, turnover, and training (Becker,
Huse[id, & Ulrich, 2001). The second is that the gap behveen maximum and
typical perfonnance is small; that is, the employee is motivated to do the job
to near the best of his/her ability most of the time. The utility analysis
approach docs not break down influences on perfonnance. such as motivation
level, interpersonal problems with other organization members, or lack of
resources needed to perfonn the job, all of which can affect an employee s per-
fonnance level. We emphasize that we do not intend our use ofthe term under-
perfonnancc to be value laden nor do we intend to use it in a pejorative sense.

CREPID and the Costs of Underperformance

To operationalize CREPID, we used a portion ofajob, programmer analyst_
which underwent job analysis under our supervision. The job analysis provided
a list of the job's essential tasks. Based on available salary data for programmer
analysts, we used an annual salary of$50,000 USD. We assigned a proportion-
al amount of the annual salary to each task and then applied hypothetical job
perfonnance ratings to each task. We then translated the resulting ratings into

The Industrial-Organizational Psychologist 15

estimates of dollar value for each task. "The sum of the dollar values assigned
to each task equals the economic value of each employee's job performance to
the company'' (Cascio, 1991, p. 214). We reasoned that the cost ofunderper-
formance is the difference, in dollars, between a high-performing employee and
low-performing employee (a conservative estimate of underperformance cost)
or between a high-performing employee and meets-expectations-performing
employee ( very conservative estimate of underperformance cost).
In applying CREPrD to this job. we followed seven steps to calculate dol-
lar amounts that would be assigned to each task for a hypothetically low-per-
forming employee, a hypothetically meets-expectations-performing employ-
ee, and a hypothetically high-performing employee. The columns within
Table 1 depict the following process:
Step J: ldenr[fy key Job tasks. The five most important tasks were identi-
fied from the job analysis (column I).
Step 2: Obtain frequency and importance ratings for each job task. Six
subject-matter experts (SMEs) assigned ratings for each task in terms of fre-
quency and importance to the job (columns 2 and 3).
Step 3: Calculate relative weight of each job task. The numerical ratings
for frequency and importance were multiplied together and divided by 65 (the
total frequency x importance value) to obtain a relative weight for each task
(column 5).
Step 4: Assign dollar value to each task. The relative weight percentage
for e.ach task was multiplied by a salary of $50.000 to obtain the dollar value
of each task {column 6).
Step 5: Rate employee pe1formance 011 each task. Performance levels for
low, meets expectations. and high performers were simulated for each task
(columns 7, 9. and 11, respectively).
Step 6: Tie together dollars and pe,.formance. The dollar value of each
task was multiplied by the performance levels of the low, meets expectations,
and high performers (columns 8, 10, and 12, respectively).
Step 7: Compute economic value ofjob pe,formance. The dollar values of
each of the five tasks were added to create the economic value of job perform-
ance for low, meets expectations, and high performers (bottom row of Table 1).
Based on our calculations, !ow-performing computer programmers who
earn a yearly salary of $50.000 have a value of $29,806.25 to the organiza-
tion and are overpaid by $20. 193.75. Computer programmers who meet
expectations have the value of exactly what they are paid. $50,000. Finally,
high-perfonning computer programmers have a value of $7 1,933.75 to the
organization and arc worth $21,933.75 more than they are paid.
To detennine the dollar amount assigned to differences in performance.
we subtracted the low-performing employee's net dollar value from the high-
performing employee's net dollar value: $71.933. 75-$29,806.25 =
$42,127.50 (see last row in Table 1). We multiplied this value by 200 as an
16 April 2011 Volume 4!l Number 4
;!
n>
5' Table I
0.
5; The Estimated Cost of Underpe,formance Among Programmer Analysts in an Organization of J,000 Workers
[ 7 8 9 IO 11
~..
::, 5 6
Rated
perf.
Rated
perf.
Rated pcrf. Rated perf. Rated
(meets (meets pert: 12
~- Relative Dollar (Low): (low): cxpecta- ellpccta- (high): Rated pcrf.
o
::,
2 3 4 weight value Pts. Net tion): Pts. tion): Net Pts. (high): Net
~ Tasks Imp Freq Total _{%) _ . ($50k) _ assigned value assigned value assigned value
~
- ---- ~
'< I. Analyze customer requirements to 5 3 15 23.1 $11,550 0.5 $5,775 I $11,550 I. 75 $20,212.50
n
::r ensure that they are met by the software
f
0

2. Analyze software requirements to deter- 5

mine feasibility of design within time
3 15 23. l $11,550 0.75 $8,663 $11,550 1.35 $15,592.50

and cost constraints.

3. Evaluate options for required reporting 5 3 15 23.l $ll,550 0.5 $5,775 511,550 1.5 $17,325.00
fonnats, cost constraints, and need for
security restrictions.
4. Define candidate architectures for evalu- 5 2 IO 15.35 $7,675 0. 75 $5,756 $7,675 $7,675.00
ation that meet critical architectural
requirements of the system
5. Define design cl!L~Ses that implement 5 2 10 15.35 $7,675 0.5 $3,838 $7,675 1.45 $11,128.75
required system capabilities
Total 65 100.00 $50,000 $29.806 $50,000 $71.933.75
estimate of the number of employees, out of 1,000 (20% ), who might be con-
sidered low performers: $42,127.50 x 200 low performers = $8,425,000. In
other words $8,425,500 might be the cost of underperfonnance in an organi-
zation of 1,000 employees.
A more conservative estimate of underperformance compares low-per-
fonning employees to employees who meet expectations (see last row in
Table l): $50.000-$29,806.25 = $20,193.75. If 80% of the employees meet
expectations and 20% are low performers, this more conservative estimate of
the cost ofunderperformance is $4,038,750 ($20,193.75 x 200).
The costs of removing underperformance from the workplace include
training costs, a supervisor's time devoted to providing assistance, instruction
and monitoring (e.g., coaching), and recruiting and selection costs. An esti-
mate of the turnover costs for a computer programmer in 2002 was $20,080,
including separation costs, replacement costs, and training costs (Mercer,
2002. as reported in Bohlander & Snell, 2007). The $20,080 figure is $113. 75
less than the cost of underperformance between a meets-expectations
employee and a low-performing employee ($50,000-$29,806.25 =
$20,193.75). Even such a seemingly small difference of$113.75, when mul-
tiplied by our estimated number of 200 low performers, is not insubstantial:
$22,750. The organization we studied would gain $22,750 by terminating
low performers and replacing them. Multiple tenninations at the same time
would increase the savings even more as some costs would be consolidated.
Assuming that the low performers are replaced by meets-expectations per
formers, the cost of turnover is recovered quickly. Obviously, replacement
costs differ depending upon location and labor market conditions. Clearly,
however, decision makers would benefit from calculating turnover costs in
order to take the most strategically sound step to address underperfonnance.
We emphasize that we are not considering a number of issues, including
underperformance over time (i.e., the cost of replacement is likely to be well
below the cost of underperfonnance across a 2-year time span and obviously
will grow over time). In addition, coworkers and supervisors who must deal
with ongoing underperforming incumbents and later with new hires will
experience productivity losses. Of course, supervisor dealings with ongoing
undcrperforming incumbents are likely more expensive than dealing with
new hires, considering time and effort devoted to attempts to raise the effec-
tiveness level of the underperfonners.

Conclusion

Asch ( 1940) argued that influence is not a change in attitude toward an

object but rather a change in the definition and meaning of the object. We
respectfully invite our colleagues to consider the possibility that focusing on
threats in tenns of"underperfonnance" rather than long-tenn opportunities to
"add value" is likely to position us to effectively sell our services to managers.

18 April 2011 Volume 48 Number 4

 References
Asch, S. E. (1940). Studies in the principles of judgments and attitudes II. Determination of
judgments by group and by ego standards. Journal of Social Psycho/og,\ 12, 433-465.
Becker, 8. E.. Huselid, M.A., & Ulrich, D. (2001). Tire HR .rcomcard. Boston. MA: Har-
vard Business School Press.
Bohlander, G, & Snell, S. (2007). Managing human resources. Mason, OH: Thomson Cor-
poration.
Boudreau, J. W. (1991 ). Utility analysis for decisions in human resource management. In M.
D. Dunnette & L. M. Hough (Eds.). Handhook of indusrrial a11d organi:atirmal p.~1'Chology (pp.
621- 745). Palo Alto, CA: Consulting Psychologists Press, Inc.
Ca~on, K. P., Becker. J. S .. & Hender.;on, J. A. ( 1998). Is utility really futile? A failure to
replicate and an extension. Joumal ofApplied Psychology, 83. 84-96.
Cascio, W. (1991). Costing human re.murces: The financial impact o.fhelra,-ior i11 organi-
:.arions (3rd edition). Boston, MA: PWS-Kent.
Cascio, W., & Ramos, R. ( 1986). Development and application of a new method for asse!IS-
ing job performance in behavioral/economic terms. Journal ofApplied P.<rcho/ogi\ 71, 20-28.
Latham. Ci P. & Whyte, G (1994). l11e futility of utility analysis. Per.<onnel Psychology,
47, 31-46.
Mecan. T. H., & Foster, L. (2004). Managers' reactions to utility analysis and perceptions of
what influences their decisions. Journal of Busb1es.< and P.")clrofog,; 19. 24 1- 253.
Mercer, M. (2002). 1itming your h11ma11 re.<m1rce.r department imo a profir center. Barring-
ton, IL: Castlegatc Publishers, Inc.
Prentice-Dunn, S., Jones, J. L., & Floyd, D. L. ( I 997). Persuasive appeals and the reduction
of skin cancer risk: The roles of appearance concern, perceived benefits of a tan, and efficacy
information. Journal ofApplied Social P.<_w:holoJ,."Y. 27, 1041 - 1047.
Rogers, R. W. ( 1983 ). Cognitive and physiological processes in fear appeals and attitude
change: A revised theory of protection motivation. In J. T. Cncioppo & R. E. Petty (Eds.), Social
psychophysiology: A sourcebook (pp. 153-177). New York, NY: Guilford.
Whyte, G. & Latham, G ( 1997). The futility of utility analysis revisited: When even an
expert fails. Personnel Psychology, 50, 601---610.
Wood, W. (2000). Attitude change: Persuasion and social influence. .411n11a/ Review of P.~v-
chologv. 51. 539--570.

The Industrial-Organizational Psychologist 19